Antiphishing model based on similarity index and neural networks

Abstract

Phishing is a negative technique that is used to steel private and confidential information over the web. In the present work author proposed a hybrid similarity of Cosine and Soft Cosine to calculate the similarity between the user query and repository as an anti-phishing approach. The proposed work model uses a multiclass learning method called Feed Forward Back Propagation Neural Network. The model evaluation results with 100 to 3000 test files shows that the hybrid model is able to detect the phishing attack with an average precision of 71% and is highly effective.